This week, I have looked into using track-level summaries. Key and chord analysis have been of less interest to me, as I am looking at performances of the same song. Instead, I have been interested in investigating Dark Star further, as it is the track that is most distinct from the rest. According to the plot, the most consistent period is the period from 1971-1972, when Dark Star as a lenghty jam song became solidified. The 1967-1968 period, too, when Dark Star was released as a single, is also consistent. The transitional period 1969-1970 period is more inconsistent - while it could be argued that this was the period where the song form was in transition, it also features the greatest deviation. This is likely due to a San Francisco performance in August 1969, where three Dark Star excerpts were played between different songs as separate tracks - the latter two being upbeat and fast.
The later periods, in comparison, are noticeably faster, with the final period marked by the highest deviations. In the last period of the Grateful Dead, Dark Star is played erratically both in form and style, being played in up to 4 different parts of the same performance and in a noticeably more experimental manner from previous periods.
While the 1969-1972 period feature the lengthiest tracks, I am not sure how much can be said about the loudness, as the recordings are often fan-taped - meaning that it is hard to use it as a variable for the music itself.
The Grateful Dead was an American jam band established in 1965 that fused many genres of music such as folk, jazz, blues, psychedelic rock, and gospel. They were especially known for their live performances with lengthy instrumental jams, which would usually be elaborations on some basis tunes - for example, there exist 80 versions alone of Dark Star on Spotify. Have these basis tunes varied in style throughout their 30-year stint - could they be split into specific “eras”?
To support my corpus, I will look at several songs that have been heavily played by the Grateful Dead throughout their career, in this way comparing and contrasting my findings. Another reason for picking various songs is also simply because they would not use some tracks at all in some periods - Dark Star was never played at all between 1974 and 1978. The basis tunes I will use will be Dark Star, Not Fade Away, China Cat Sunflower, and Me And My Uncle, all songs the Grateful Dead have played around 400-600 times each in their career.
I begin my preliminary questions by asking - can we even see differences of performances over time? I would argue yes, as, for example, overall valence has clearly gone down for both China Cat Sunflower, Me and My Uncle, and Not Fade Away, with valence generally peaking in the first half of the 1970s.
An interesting odd-one-out is Dark Star. Dark Star was performed since the foundation of the Grateful Dead, but the song was initially released as an upbeat single, which is reflected in the higher valence in their early years. It was the highly-acclaimed Live/Dead live album, recorded in 1969, that signalled the song’s evolution into the a cornerstone of the band’s jamming. Another variable that can help us see how the song became the tune of choice for jamming on is the track duration - the last chart shows how the length of the Dark Star tracks is inversely correlated with the valence, showing how the song evolved from being upbeat to drawn-out, before slightly reversing the trend back to upbeatness in their later years.
In order to spot differences between different recordings of the same song, I created a plot that shows dynamic time warping between two recordings. To begin my investigations, I attempted to compare two recordings of Me And My Uncle, a relatively simple cover song without much improvisation, and chose two recordings with virtually the same length and similar recording dates. The fact that two of the most similar-at-appearance tracks do not seem to converge on the graph complicates the usage of using DTW for my corpus.
In the process